This invention relates to an optical modulation apparatus and to a method of controlling an optical modulator. More particularly, the invention relates to an optical modulation apparatus and to a method of controlling an optical modulator, wherein even if the operating point of an optical modulator the optical output of which varies periodically with respect to a driving voltage fluctuates owing to a change in ambient temperature or aging, the fluctuation in operating point can be compensated for in stable fashion. More specifically, the present invention relates to a control method for stabilizing the operating point of a Mach-Zehnder optical modulator (referred to as an "MZ-type optical modulator) in an optical transmitter used in a time-division multiplexing (TDM) or wavelength-division multiplexing (WDM) optical transmission system.
The explosive increase in the quantity of available information in recent years has made it desirable to enlarge the capacity and lengthen the distance of optical communications systems. In-line optical amplifier systems which accommodate a transmission speed of 10 Gbps are now being put to practical use. Even greater capacity will be required in the future, and research and development is proceeding in both the TDM and WDM aspects of optical transmission.
Direct Modulation
Intensity modulation and direct detection (so-called "direct modulation") is the simplest technique to use for an electro-optic conversion circuit employed in an optical communications system. According to this technique, a current that activates a semiconductor laser is turned on and off directly by the "0"s and "1"s of a data signal to control the emission and extinction of the laser beam. When a laser per se is turned on and off directly, however, the light signal experiences a fluctuation in wavelength (so-called "chirping") owing to the properties of the semiconductor. The higher the transmission speed (bit rate) of the data, the greater the influence of chirping. The reason for this is that an optical fiber exhibits a chromatic dispersion property wherein propagation velocity varies for different wavelengths. When chirping is caused by direct modulation, propagation velocity fluctuates, waveforms are distorted during propagation through optical fiber and it becomes difficult to perform long-distance transmission and transmission at high speed.
External Modulation
For the reasons mentioned above, external modulation is used for high transmission speeds of 2.5 to 10 Gbps. According to external modulation, a laser diode emits light continuously and the emitted light is turned on and off by the "1"s and "0"s of data using an external modulator. The above-mentioned MZ-type modulator primarily is used as the external modulator. FIGS. 32A and 32B are diagrams useful in describing the MZ-type modulator, in which FIG. 32A is a schematic view of the construction of the modulator and FIG. 32B is for describing the modulating operation.
Shown in FIG. 32A are a distributed-feedback semiconductor laser diode (DFB-LD) 1 used in long-distance transmission at a speed of greater than 1 Gbps, an MZ-type modulator 2 and optical fibers 3a, 3b. The MZ-type modulator 2 includes on an LiNbO.sub.3 substrate, (1) an input optical waveguide 2a formed on the substrate for introducing light from the laser diode 1, (2) branching optical waveguides 2b, 2c and (3) an output optical waveguide 2d formed on the substrate for outputting modulated light; (4) two signal electrodes 2e, 2f formed on the substrate for applying phase modulation to the optical signals in the branching optical waveguides 2b, 2c, and (5) a signal input terminal 2g formed on the substrate for inputting an NRZ electrical drive signal to one of the signal electrodes, namely the electrode 2e.
If a voltage applied to the signal electrodes 2e, 2f is controlled by the "1"s and "0"s of data, the branching optical waveguides 2b, 2c develop a difference in refractive index and the light waves of the optical signals in the optical waveguides develop a difference in phase between them. For example, if the data is a "0", control is performed in such a manner that the phase difference between the light waves of the optical signals in the two optical waveguides 2b, 2c becomes 180.degree.; if the data is a "1", control is performed in such a manner that the phase difference between the light waves of the optical signals in the two optical waveguides 2b, 2c becomes 0.degree.. If this arrangement is adopted, superimposing the optical signals of the two optical waveguides 2b, 2c will make it possible to output the input light upon modulating it (turning it on and off) by the "1"s and "0"s of the data.
As shown in FIG. 32B, the optical output characteristic of the MZ-type optical modulator, which has a voltage difference between the two electrodes thereof, varies periodically in dependence upon the applied voltage. Point A represents the culmination of the light emission and point B the culmination of extinction. The range of the voltage over one period is 2V.pi.. When data is a "1", therefore, voltage having an amplitude of V.pi. is applied between the signal electrodes 2e, 2f, whereby light is emitted. When data is a "0", a voltage of zero is applied between the signal electrodes 2e, 2f, whereby light is extinguished.
The MZ-type optical modulator described above is advantageous in that transmitted light exhibits little chirping. However, a change in the temperature of the LiNbO.sub.3 constituting the substrate, prolonged application of an electric field thereto and aging thereof are accompanied by polarization of the substrate per se, electric charge remains on the surface of the substrate and the bias voltage across the signal electrodes fluctuates. Consequently, the voltage--optical output characteristic of the MZ-type optical modulator fluctuates to the left and right from the ideal curve a in FIG. 33 to the curves b and c. In other words, the operating point of the MZ-type optical modulator drifts with the passage of time, thereby the on/off light level changes and causes inter symbol interference between codes (refer to output eye patter in FIG. 33).
Bias Control Method in NRZ Modulation
Accordingly, in order to stabilize the operating point, the conventional practice is to perform control in such a manner that the bias voltage is increased correspondingly if the curve shifts to the right and decreased correspondingly if the curve shifts to the left. More specifically, there has been proposed a compensation method (referred to as "automatic bias-voltage control" (ABC) below) which includes superimposing a low-frequency signal on an electrical drive signal, detecting the amount of drift of the operating point and the direction of this drift, and controlling the bias voltage by feedback (see the specification of Japanese Patent Application Laid-Open No. 3-251815). FIG. 34 is a diagram showing the construction of a circuit for stabilizing the operating point of an optical modulator that implements the currently available method of compensating the modulator operating point, and FIG. 35 is a diagram useful in describing the principle of operating-point stabilization.
Shown in FIG. 34 are the semiconductor laser diode (DFB-LD) 1, the MZ-type optical modulator (LN optical modulator) 2, the optical fibers 3a, 3b and a drive circuit 4. An NRZ electric signal (the data signal) is input to the drive circuit 4, which proceeds to generate an electrical drive signal SD having an amplitude (=V.pi.) between the culmination A of light emission and the culmination B of light extinction in the voltage optical output characteristic (see FIG. 32B) of the MZ-type optical modulator 2. A low-frequency oscillator 5 generates a low-frequency signal SLF having a low frequency f.sub.0 (e.g., 1 KHz), a low-frequency superimposing circuit 6 for superimposes a low-frequency signal on the drive signal SD, an optical branching unit 7 branches the optical signal from the optical modulator 2, and a light receiver (PD) 8 such as a photodiode converts the optical signal output by the optical modulator 2 to an electrical signal. Numeral 9 denotes an amplifier. A phase comparator 10 detects and outputs a phase difference .theta. between the low-frequency signal component of the frequency f.sub.0 contained in the optical signal output by the optical modulator 2 and the low-frequency signal output by the low-frequency oscillator 5. A low-pass filter (LPF) 11 rectifies the output signal of the phase comparator 10, and a bias supply circuit 12 controls the bias voltage, which is applied to a signal electrode, in such a manner that the phase difference .theta. will become zero.
The low-frequency superimposing circuit 6 subjects the drive signal of the MZ-type optical modulator 2 to amplitude modulation by the signal having the low frequency of f.sub.0, the photodiode 8 converts the output light of the optical modulator 2 to an electrical signal, the phase comparator 10 performs a phase comparison between the low-frequency signal impressed upon the drive signal and the low-frequency signal component contained in the optical signal, and the bias supply circuit 12 controls the bias voltage applied to the signal electrode in such a manner that the phase difference .theta. will become zero.
The optimum operating points of the MZ-type optical modulator are points A and B (see FIG. 35) at which the two levels of the waveform of the drive signal SD give the maximum and minimum output optical powers. In the case that there is no fluctuation in the voltage optical output characteristic of the MZ-type optical modulator 2. Even if the signal SLF having the low frequency f.sub.0 is impressed upon the drive signal SD, upper and lower envelopes ELU, ELL of the output light do not contain the f.sub.0 component and a frequency component that is twice f.sub.0 appears in the ideal state (curve a).
On the other hand, if the characteristic curve shifts to the left or right from a to b or from a to c (if the operating point shifts to the left or right) in the manner illustrated, the upper and lower envelopes ELU, ELL of the output light both become signals modulated by the same phase. These signals contain the fo component. In addition, the phases of the upper and lower envelopes ELU, ELL of the output light in characteristic curve b are the opposite of the phases of the upper and lower envelopes ELU, ELL of the output light in characteristic curve c.
By virtue of the foregoing, the direction in which the operating point drifts can be detected by comparing the phase of low-frequency signal SLF superimposed on the drive signal and the phase of the low-frequency signal component contained in the optical signal. The bias voltage can be controlled in such a manner that this phase difference will become zero.
Optical Duobinary Modulation
In a case where an increase in capacity is intended by TDM, a factor is that chromatic dispersion (GVD) governs transmission distance. Dispersion tolerance is inversely proportional to the square of the data transmission speed (the bit rate). A dispersion tolerance that is about 800 ps/nm in a 10-Gbps system, therefore, deteriorates to about 1/16 of this figure, namely to about 50 ps/nm, in a 40-Gbps system. One method of reducing waveform degradation due to chromatic dispersion is optical duobinary modulation. (For example, see A. J. Price et al., "Reduced bandwidth optical digital intensity modulation with improved chromatic dispersion tolerance", Electron. Lett., vol. 31, No. 1, pp. 58-59, 1995.)
In comparison with the NRZ modulation scheme, optical duobinary modulation reduces the bandwidth of the optical signal spectrum to about half thereby it reduces the effects of chromatic dispersion. For example, whereas the bandwidth of the optical signal spectrum of a 10-Gbps NRZ signal is 10 GHz in terms of frequency and 0.2 nm in terms of wavelength, the bandwidth of the optical signal spectrum of a 10-Gbps duobinary signal is 5 GHz in terms of frequency and 0.1 nm in terms of wavelength. Because the velocity of light differs depending upon wavelength, the larger the bandwidth of the spectrum of the optical signal, the greater the amount of change in the velocity at which light propagates and, hence, the greater the distortion of the waveform caused by long-distance transmission. Accordingly, if the bandwidth of the spectrum of the optical signal can be made small by optical duobinary modulation, the amount of fluctuation in velocity can be reduced and the dispersion tolerance can be increased.
FIG. 36 is a diagram showing the construction of a modulation apparatus that relies upon optical duobinary modulation, FIGS. 37A, 37B are diagrams useful in describing the principle of optical duobinary modulation, and FIGS. 39A, 39B are waveform diagrams of the associated signals.
Shown in FIG. 36 are the semiconductor laser diode (DFB-LD) 1 and the MZ-type optical modulator 2 having two signal electrodes for applying phase modulation to the optical signals in the optical waveguides on both sides, and drive-signal input terminals for inputting complimentary drive signals to the signal electrodes.
A precoder 21 encodes a 40-Gbps binary NRZ electrical input signal. A D-type flip-flop (D-FF) 22 extracts and stores the output of the precoder 21 at a 40-GHz clock and outputs a non-inverted signal D and an inverted signal *D. Phase shifters 23a, 23b adjust the output phases of the flip-flop 22 and apply there outputs to amplitude adjusters 24a, 24b, respectively. The outputs thereof are applied to electrical low-pass filters 25a, 25b, respectively, having a bandwidth that is one-fourth the bit rate (=40 Gbps). Bias adjustment circuits (bias tees) are shown at 26a, 26b and terminators at 27a, 27b. The binary NRZ electrical input signal encoded by the precoder 21 is made 3-value electrical signals S1 and S2 having inverted signs by passage through the low-pass filters 25a, 25b, and these signals are in turn passed through the bias tees 26a, 26b, thereby generating complimentary 3-value electrical drive signals (push-pull signals) S1', S2' that are applied to the respective ones of the two symmetrical signal electrodes of the MZ-type optical modulator 2.
In the MZ-type optical modulator 2, the driving amplitude necessary to turn the CW light on and off generally is V.pi. (see FIG. 37B) based upon the voltage optical output characteristic. In optical duobinary modulation, however, each of the two signal electrodes is subjected to push-pull modulation by the amplitude Va. (This is modulation in which voltages that are always opposite in sign are applied to the two electrodes). The voltage applied to the optical modulator 2 is the voltage difference (=S1'-S2') between the input signals S1' and S2'. In optical duobinary modulation, in other words, the MZ-type optical modulator 2 is modulated by a driving amplitude 2V.pi., namely an amplitude that is twice V.pi.. Further, the bias voltage (the center voltage of the electrical signal) is set in such a manner that the optical modulator is driven between two light-emission culminations A, A on the voltage--optical output characteristic curve.
The details of optical duobinary modulation will now be described.
As shown in FIG. 38, the precoder 21 includes a NOT gate 21a for inverting an input signal an, a 1-bit (25 ps) delay gate 21b, and an EX-OR gate 21c for outputting a signal c.sub.n obtained by taking the exclusive-OR between the preceding output c.sub.n-1 and the present inverted input b.sub.n. If reference is had to a truth table of the inverted signal b.sub.n, the preceding output signal c.sub.n-1 of the EX-OR gate and the present output signal c.sub.n of the EX-OR gate, we have the following:
(1) c.sub.n =c.sub.n-1 (no change in sign) if b.sub.n ="0" holds; and PA1 (2) c.sub.n =1-c.sub.n-1 (sign inverted) if b.sub.n ="1" holds. PA1 (3) outputs the 0.0 level in a case where the output c.sub.n of the precoder is successive "0"s ("00": no change in sign); PA1 (4) outputs the 1.0 level in a case where the output c.sub.n of the precoder is successive "1"s ("11": no change in sign); and PA1 (5) outputs the 0.5 level in a case where the sign of the output c.sub.n of the precoder reverses ("01 or "10"). PA1 (1) "1" is output (light is emitted) if the input data an is "1", at which value 2V.pi. or 0 is input across the signal electrodes of the MZ-type optical modulator 2; and PA1 (2) "0" is output (light is extinguished) if the input data an is "0", at which value V.pi. is input across the signal electrodes of the MZ-type optical modulator 2. PA1 (1) superimposing the low-frequency signal on the drive signal in such a manner that phases of upper and lower envelopes of the drive signal coincide; PA1 (2) superimposing the low-frequency signal on the drive signal in such a manner that only an upper or a lower envelope of the drive signal varies; PA1 (3) superimposing the low-frequency signal on the drive signal in such a manner that amplitudes of upper and lower envelopes of the drive signal differ; PA1 (4) superimposing the low-frequency signal on the drive signal in such a manner that frequencies of upper and lower envelopes of the drive signal differ; and PA1 (5) superimposing the low-frequency signal on the drive signal in such a manner that phases of upper and lower envelopes of the drive signal differ.
A low-pass filter 25a has a bandwidth which is only one-fourth of the bit rate, namely 10 GHz. Consider two successive bits of the input signal cn. If the input data varies at high speed in the manner "0, 1" or "1, 0", the low-pass filter 25a cannot follow up this change and outputs 0.5, which is the level intermediate the 0 and 1 levels. If the input data is two successive "1"s, namely "1, 1", the low-pass filter 25a outputs the level 1.0; if the input data is two successive "0"s, namely "0, 0", the low-pass filter 25a outputs the level 0.0. More specifically, the low-pass filter 25a:
From (1) to (5) above, the output of the low-pass filter 25a changes if the sign of the precoder output changes. That is, the low-pass filter 25a outputs the 0.0 or +1.0 level as the output d.sub.n if the input data an is "1", and outputs the +0.5 level as the output d.sub.n if the input data an is "0". Similarly, the low-pass filter 25b outputs the 0.0 or -1.0 level as the output *d.sub.n if the input data an is "1", and outputs the 0.5 level as the output *d.sub.n if the input data an is "0". Accordingly, if the level .+-.1.0 is .+-.V.pi. and the level .+-.0.5 is .+-.V.pi./2, then 2V.pi. or 0 is input across the signal electrodes of the MZ-type optical modulator 2 when the input data an is "1" and V.pi. is input across the signal electrodes of the MZ-type optical modulator 2 when the input data an is "0". As a result, with reference to FIG. 37B,
Thus, the waveforms of the output signals S1, S2 from the low-pass filters 25a, 25b are as shown in FIG. 39A, and the optical signal output S3 from the MZ-type optical modulator 2 becomes as shown in FIG. 39B.
The characterizing feature of the optical duobinary modulation method is that the bandwidth of the optical signal spectrum is approximately half that obtained with the conventional NRZ modulation method described above. This makes it possible to reduce the effects of chromatic dispersion.
Further, in accordance with optical duobinary modulation, channels can be disposed at higher density in the WDM scheme. In a case where the intent is to enlarge capacity by the WDM technique, bandwidth of wavelength at which a optical amplifier can amplify are limiting factors. However, if optical duobinary modulation is used, the fact that this method provides a narrow bandwidth for the optical signal spectrum can be utilized and channels can be disposed at a higher density within the amplification bandwidth of the light amplifier.
Further, in optical duobinary modulation, chirping can be reduced because of push-pull drive. Chirping occurs and the direction thereof reverses when the applied voltage of an optical modulator increases and decreases. With optical duobinary modulation, however, the electrodes are driven by mutually complimentary electrical signals. Consequently, when the applied voltage increases at one electrode, it decreases at the other, and when the applied voltage decreases at one electrode, it increases at the other. Since the optical phase of the output optical signal is the sum of the optical phases produced at the two electrodes, chirping is reduced by cancellation.
An advantage of the MZ-type optical modulator is the fact that transmitted light experiences little chirping, as mentioned above. However, a change in the temperature of the LiNbO.sub.3 constituting the substrate and the aging thereof are accompanied by temporal drift of the operating point of the voltage--optical output characteristic.
For this reason, it is necessary to control the bias voltage in dependence upon drift of the operating point, just is in the NRZ modulation scheme, in optical duobinary modulation as well. However, the problems set forth below arise when the operating-point compensation technique of NRZ modulation is applied directly to optical duobinary modulation. FIG. 40 is a diagram useful in describing a case where the operating-point compensation technique of NRZ modulation is applied directly to optical duobinary modulation.
With optical duobinary modulation, the driving voltage is made twice that used in NRZ modulation. Consequently, if the voltage--optical output characteristic shifts to the left or right from the ideal characteristic a to b or c, the envelopes ELU, ELL of the optical signal corresponding to the ON-side and OFF-side portions EU and EL of the electrical driving signal of the modulator subjected to low-frequency modulation take on mutually opposite phases and cancel each other out, making it impossible to detect the signal component of the low frequency f.sub.0. The problem that arises, therefore, is that the ABC control method employed in the conventional NRZ modulation method cannot be applied to a modulation scheme, which includes optical duobinary modulation, wherein an optical modulator is driven between two light-emission culminations or between two light-extinction culminations of the voltage--optical output characteristic.
Another problem is that the conventional ABC control method only assumes use of an MZ-type optical modulator configured for electrode drive on one side. This means that it is necessary to also consider setting of an operating point in a case where an optical modulator configured for driving electrodes on both sides is used in optical duobinary modulation, NRZ modulation and RZ modulation.